Chess Board - Piece Recognition + Robot Control

Hey guys

I am going for a big project including a robot which is moving chess pieces and can play with other people. The robot will move the pieces from underneath with an electro-magnet.

My goal is to detect when a piece is on a field and which piece exactly it is!

The last days I read a lot about past projects and I think this will be quite a challenge..

I have thought about RFID/NFC but it is difficult to implement (isolation) and I also think the magnets would disturb the signal..

I've also thought about Barcode, Color Detection, Hall-effect, Reed Sensors... but I don't really find an ideal solution.

Before I start frustrating and keep searching for more solutions, maybe someone of you has a good idea?

I have never seen a project with strain gauge. What do you think about that idea? Putting strain gauges underneath each field and modify the pieces in weight.

(I don't really like to use cameras)

why not?
It's 2023 and object recognition out of a video stream will work very precise.

Knights horse jumps..
Castling be tricky too..

sorry.. ~q

The fields just have to be big enough so the horse can go around. That's not a problem.

Thanks! That's a nice project but there is no individual piece recognition.

The location identifies the piece. Computers have memory.

Putting strain gauges underneath each field and modify the pieces in weight.

With a bit of research, you will understand why this idea is unworkable.

I know that computers have memory but that is not my aim. I want to identify each piece individually.

Your aim is perfectly clear. Forum members are telling you that your current ideas will not work, or are completely impractical to implement.

Maglev! : )

(or, every piece move over a little to let the horse pass)

1 Like

Ok thanks. Then I come to the conclusion that I will have to detect the presence of the pieces with reed or hall-effect sensors and additional I can implement computer vision for identification of each piece. Hmm

I have experimented a bit during the last days. What do you guys think about the idea to color the individual pieces from the bottom and you leave a small "window" where the camera can detect from below which figure exactly it is.

I think this is easier and faster for the recognition?

This topic was automatically closed 180 days after the last reply. New replies are no longer allowed.